A comparative analysis of database connection pooling implementations with emphasis on the added value of aspect orientation

Abstract

Enterprise web based systems require an extremely vigilant management of database resources. Database connectivity is considered to be an important crosscutting, nonfunctional requirement that inevitably impacts the overall, runtime performance measures of the running web application. It is also a requirement that needs systematic resolution at the early phases of designing the web application under development due to its crosscutting nature. Database connection pooling can be seen as one of the more efficient solutions that will enhance the run-time performance of web/database transactional activities. Our work asserts that adopting an aspect-oriented approach for database connection pooling in web based systems promises for an even better improved run time performance of web based systems. In this study we analyze the run-time effects that result from applying three approaches upon cross cutting concerns of database connectivity in a developed web application. The three approaches are: An aspect-oriented approach (AOP), an object-oriented approach (OOP) or (Pool), and sequential calls to the native database driver approach (Native). The run-time effects were translated into run-time performance measures, namely, the average response time and successful hits percentage. We were able to demonstrate the added benefit of using an aspect-oriented approach for this problem.